## wpiea2022098-print-pdf

## Source details

**Canonical URL:** [wpiea2022098-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022098-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022098-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022098-print-pdf.pdf.json)

---

### Introduction and context
- Eurobonds provide international diversification, relative liquidity, transparency, and flexible access to a wide investor base for emerging market and developing country (EMDC) sovereigns.
- Market scale and composition (as reported):
  - By 2020 there were about US$1.5 trillion in rated sovereign EMDC Eurobonds outstanding—up from under US$1 trillion in 2015.
  - About 70 new issues per year over the past decade.
  - As of 2017, 82 countries had issued Eurobonds.
  - About US$500 billion in non-sovereign Eurobonds were outstanding (issued by major EMDC banks and enterprises).
- Eurobond characteristics: issued in a jurisdiction (typically New York or England) and a currency (predominantly the US dollar) distinct from those of the issuer; sold “over the counter” and recorded in major international central securities depositories.
- Importance: Eurobond pricing is monitored as an indicator of market perception of risk and used as a benchmark for other products.

### Liquidity concept and connection to yields
- Market liquidity defined: ability to buy or sell a reasonable amount of a security, reasonably quickly, with minimal effect on price.
- Bid-ask spread is a necessary but not sufficient liquidity metric; what is “modest” varies by market (a few basis points in advanced wholesale bond markets; some tens of basis points in emerging market bond markets).
- Liquidity value:
  - Facilitates price discovery and enables investors to adjust positions quickly.
  - Eases debt management operations by DMOs (selling more securities; liability management operations).
  - Critical during crisis periods: liquidity disappearance can amplify shocks and contribute to market freeze or systemic collapse.
- Hypothesis: bond characteristics that promote liquidity (lower bid-ask spreads) should be rewarded by lower yields.

### Theoretical determinants of bid-ask spreads and liquidity
- Drivers of dealer bid-ask spreads:
  - Operational costs (including fixed elements) → spreads inversely related to turnover.
  - Dealer market power.
  - Financing and capital costs for inventory; security riskiness increases spread.
  - Adverse selection risk.
- Implications: spreads should relate to turnover, investor variety, issuance volume, credit and market risk, duration, and standardization; feedback loops exist between liquidity and yields.

### Empirical evidence — domestic and international findings (summary)
- General: yields tend to increase with illiquidity; common market-wide and issuer-specific liquidity factors matter, especially in stress times.
- International / Eurobond-specific findings:
  - Higher bid-ask spreads associated with higher yields; small issue size associated with higher yields.
  - Hund and Lesmond (2008): liquidity significant for sovereign and corporate emerging market bonds; bid-ask spread correlates with other liquidity measures and remains significant after joint estimation and instrumental approaches.
  - Term-and-condition features (CACs, jurisdiction) affect yields and liquidity; enhanced CACs generally reduce holdout risk and can lower yields.

### Determinants of bid-ask spreads — empirical findings
- Credit ratings:
  - Initial and current credit ratings both have economically significant effects; change in rating since issuance has a strong effect.
  - Initial rating matters most for countries with below-average credit ratings.
- Coupon rate:
  - Distinctly positive but nonlinear relationship with the bid-ask spread.
  - A coupon rate of 1.5 percent is associated with a bid-ask spread about 25 basis points lower than in the case of a coupon rate of 5 percent.
- Inclusion in the EMBIG:
  - Small effect on bid-ask spread; positive for low-rated issuers and negative for medium- to high-rated issuers; estimated parameters not individually significant but jointly significant.
- Issuance jurisdiction:
  - Issuance under New York increases bid-ask spreads for low-rated bonds; no significant effect for medium- to high-rated bonds.
- Contractual clauses:
  - Original CAC: only a slight negative (insignificant) effect on spreads.
  - Enhanced CAC: highly significant and narrows bid-ask spreads, especially for low-rated bonds (about 20 basis points reduction for very low ratings vs. sample average spread of 55 basis points; about 5 basis points for medium-rated bonds).
- Issue volume and vintages:
  - Increasing issue volume has an overall negative effect on spreads; effect reduced to about 20 basis points when country dummies included.
  - Deviation from country average issue volume tends to increase spreads.
  - Issue year effects: 1999–2003 show unusually large spreads; parameters stable from 2004 onward.
- Predictive regressions using only variables known at issuance:
  - R^2 just over 0.5 (about 18 percentage points lower than full explanatory set).
  - Issue volume major influence for issues < US$1 billion, with threshold effects at US$500 million and higher.
  - Seasoning: liquidity predicted to dry up in final few years before maturity.

### Determinants of yields — empirical findings
- Explanatory power:
  - Regressions for log yields achieve R^2 around 0.85 using all explanatory variables.
- Issue volume:
  - Important, especially for mid-range ratings and volumes below US$1 billion.
  - Example stated effect: reducing volume from US$1.5 billion to US$750 billion raises the yield by over 60 basis points for a mid-rating bond.
  - Threshold effects most pronounced at US$500 million.
- Remaining volume outstanding:
  - Modest effect until remaining volume falls below a third or a quarter of initial amount, then rates rise steeply.
- Total country volume:
  - For most countries yields tend to increase with total amount outstanding; effect modest and often insignificant. For smaller, lower-rated issuers aggregate volume increases can markedly decrease yields.
- Maturity and seasoning:
  - Remaining maturity affects yield monotonically; seasoning non-monotonic (for a 15-year initial term, yield falls for first decade then rises sharply as maturity approaches).
- Credit rating and coupon:
  - Lower credit rating associated with higher yield; current rating matters most.
  - Low coupon rate associated with low yields; curvature sharp above about 4.5 percent for most rating levels.
- Contractual clauses and jurisdiction:
  - Inclusion in EMBIG has significant but modest overall effect on yields.
  - Issuance under New York law raises yields markedly, especially for low-rated bonds.
  - Original CAC: small negative effect on yields (jointly significant).
  - Enhanced CAC: markedly reduces yields, especially on low-rated bonds—depending on rating, yields can be reduced by tens of basis points.
- Predicting yields using only issuance-time variables:
  - Excluding contemporaneous and fixed effects reduces R^2 on LYLD_EMD to around 0.6.
  - Initial issue volume remains a major predictor; threshold effects at US$500 million stronger when using issuance-time variables.

### Interaction of bid-ask spreads and yields — main quantitative result
- Approach: orthogonalize yields and spreads to all explanatory variables and regress yield residuals on bid-ask-spread-related residuals to address endogeneity modestly (instrumental approach not feasible).
- Main quantitative result:
  - An increase (decrease) in the bid-ask spread of 10 basis points, starting from the sample mean of 55 basis points, increases (decreases) the yield by about 20 basis points for a typical bond with a mid-range credit rating.
- Comparable literature:
  - Hund and Lesmond: a 1 bp increase in the bid-ask spread raises corporate bond yield spreads by 3.13 bps and government bonds by 5.67 bps after controls.
  - Chen et al.: a 1 bp increase in the bid-ask spread of US corporate bonds increases yield spreads by 0.42 bps for investment grade bonds.

### Box — CACs and Pari-Passu Clauses (summary)
- Purpose: contractual clauses facilitate restructuring and reduce value destruction from disorderly restructurings; important given absence of sovereign insolvency procedure.
- Collective Action Clauses (CACs):
  - Enable qualified majority to bind minority; original CACs operated series-by-series; enhanced CACs allow two-limb aggregation or single-limb aggregation with uniformity conditions.
  - Empirical evidence: enhanced CACs reduce holdout risk.
- Pari-passu clauses:
  - Protect creditors from legal subordination; modified pari-passu clauses make explicit that payments are not “ratable,” permitting servicing of some creditors while not making pro rata payments to holdouts.
- Jurisdictional variation:
  - Formulation and implications differ across jurisdictions; e.g., New York law bonds often use a trust structure affecting protections against holdouts.
- History and consensus:
  - CACs promoted by IMF and others post late-1980s/1990s crises; enhanced CACs and modified pari-passu clauses widely included in Eurobonds issued since 2015.

### Key regression evidence and statistics (selected exact figures)
- Main regression sample sizes and transformed-variable statistics:
  - Full sample (Table 3a; No. Obs. = 4,759):
    - BAS_EMD mean = 62.78; Std. Dev. = 47.49; Min = 0.10; Max = 1,329.
    - YLD_EMD mean = 531.22; Std. Dev. = 506.74; Min = 17.40; Max = 21,831.
    - VOL_ISS mean = 1.62; Std. Dev. = 1.08; Min = 0.10; Max = 7.00.
    - CRR_ISS mean = 12.30; Std. Dev. = 3.45; Min = 6; Max = 20.
    - D_EMBIG mean = 0.54; D_NY_LAW mean = 0.55; D_O_CAC mean = 0.50; D_E_CAC mean = 0.36.
  - Main regression sample transformed variables (Table 3b; No. Obs. = 4,363):
    - LBAS_EMD mean = 4.018; Std. Dev. = 0.371; Min = 3.220; Max = 5.162.
    - LYLD_EMD mean = 6.156; Std. Dev. = 0.341; Min = 4.867; Max = 7.019.
    - LVOL_ISS mean = 0.279; Std. Dev. = 0.649; Min = -2.303; Max = 1.946.
    - LCRR_ISS mean = 2.474; Std. Dev. = 0.275; Min = 1.792; Max = 2.996.
    - D_E_CAC mean = 0.383; D_NY_LAW mean = 0.547; D_O_CAC mean = 0.489; D_EMBIG mean = 0.551.
  - Sub-sample with rating > 14 (Table 3c; No. Obs. = 978):
    - LBAS_EMD mean = 3.904; LYLD_EMD mean = 5.944.
    - D_E_CAC mean = 0.442; D_O_CAC mean = 0.443.
  - Sub-sample with total country volume ≤ US$10 billion (Table 3d; No. Obs. = 1,222):
    - LBAS_EMD mean = 4.218; LYLD_EMD mean = 6.256.
    - D_E_CAC mean = 0.373; D_O_CAC mean = 0.514.
- Log Bid-Ask Spread determinants (selected reported coefficients from Table 4; full sample, reduced form examples):
  - Sample size for main regression: 4,363 observations.
  - LVOL_ISS: 6.578*** (s.e. 1.473) in one specification.
  - LVOL_ISS_LCRR_I: -5.106*** (s.e. 1.153).
  - CHG_LVOL: -25.988*** (s.e. 3.892).
  - D_E_CAC: -5.131*** (s.e. 0.872) with D_E_CAC_LCRR_I = 4.111*** (s.e. 0.709).
  - LMAT_NOW_SQ: -0.060*** (s.e. 0.013).
  - D_NY_LAW: 1.679*** (s.e. 0.582) with D_NY_LAW_LCRR_I = -1.268*** (s.e. 0.474) and D_NY_LAW_LCRR_I_SQ = 0.226** (s.e. 0.095).
  - Adjusted R2 examples across specifications: 0.610 to 0.723.
- Log Yield determinants (selected reported coefficients from Table 8; full sample, reduced form examples):
  - Full sample: 4,363 observations; Adjusted R2 often ≈ 0.873 in several specifications.
  - LVOL_ISS: 4.440*** (s.e. 0.620).
  - LVOL_ISS_LCRR_I: -3.463*** (s.e. 0.488); LVOL_ISS_LCRR_I_SQ: 0.673*** (s.e. 0.095).
  - CHG_LVOL: -11.578*** (s.e. 2.129) with CHG_LVOL_LCRR_I = 8.572*** (s.e. 1.656) and CHG_LVOL_LCRR_I_SQ = -1.612*** (s.e. 0.320).
  - D_E_CAC: -2.298*** (s.e. 0.496) with D_E_CAC_LCRR_I = 1.924*** (s.e. 0.401) and D_E_CAC_LCRR_I_SQ = -0.394*** (s.e. 0.080).
  - D_NY_LAW: 1.989*** (s.e. 0.310) with D_NY_LAW_LCRR_I = -1.650*** (s.e. 0.252) and D_NY_LAW_LCRR_I_SQ = 0.332*** (s.e. 0.051).
- Endogeneity and residuals regressions (Table 12 examples):
  - Residual coefficients example: R_LBAS = 0.182*** (s.e. 0.011) in one reported column.
  - R_LBAS_SQ = 0.101*** (s.e. 0.035) in full sample column.
  - F test for joint significance of residual-based bid-ask terms:
    - Full sample F(4, 4358) = 80.73*** (Adjusted R2 = 0.100 in that specification).
    - Sub-sample with rating >14: F(4, 3380) = 75.62*** (Adjusted R2 = 0.135).

### Main effect magnitudes (summary from Table 1; exact reported effects)
- Decrease BAS by 10 bps. requires / Decrease YLD by 30 bps. requires (contextual framing in table).
- Issue volume dummy step effects on BAS_EMD and YLD_EMD:
  - Issue volume ≥US$0.5 bn.: Effect on BAS_EMD = -9.0 bps.; Effect on YLD_EMD = -70 bps.
  - Issue volume ≥US$1.0 bn.: Effect on BAS_EMD = -5.0 bps.; Effect on YLD_EMD = -18 bps.
  - Issue volume ≥US$1.5 bn.: Effect on BAS_EMD = -3.0 bps.; Effect on YLD_EMD = -23 bps.
  - Issue volume ≥US$3.0 bn.: Effect on BAS_EMD = -0.5 bps.; Effect on YLD_EMD = +13 bps.
- Inclusion in EMBI Global:
  - Effect on BAS_EMD = -1.5 bps. when CRR=BB+; +3.0 bps. when CRR=B-.
  - Effect on YLD_EMD = -1.5 bps. when CRR=BB+; +0.5 bps. when CRR=B-.
- Issued under NY law:
  - Effect on BAS_EMD = +2.0 bps. at CRR=BB+; +11.0 bps. when CRR=B-.
  - Effect on YLD_EMD = +17.0 bps. at CRR=BB+; +67.0 bps. when CRR=B-.
- Inclusion of an original CAC:
  - Effect on BAS_EMD = -1.5 bps. at CRR=BB+; -3.0 bps. when CRR=B-.
  - Effect on YLD_EMD = -19.0 bps. at CRR=BB+; +6.5 bps. when CRR=B-.
- Inclusion of an enhanced CAC:
  - Effect on BAS_EMD = -5.5 bps. at CRR=BB+; -18.0 bps. when CRR=B-.
  - Effect on YLD_EMD = -19.5 bps. at CRR=BB+; -61.0 bps. when CRR=B-.
- Credit rating effects:
  - One rating grade improvement reduces yield approximately 60 bps. (note 2/).
  - One rating grade improvement reduces yield approximately 45 bps. (note 3/).

### Robustness, diagnostics, and notable limitations
- Bid-ask spread regressions:
  - Robust regression estimates close to OLS; inclusion of country dummies increases R^2 by about 10 percentage points (to 0.77).
  - Residuals close to normal with modest heteroskedasticity; removing largest 12 residuals (~0.5 percent of sample) addresses leptokurtosis.
  - Sub-sample rating>14 residuals show asymmetric and very large deviations from normality.
- Yield regressions:
  - Results qualitatively robust; including country dummies increases R^2 by about 7 percentage points.
  - Regression for rating>BBB (=14) explains high proportion of variance but many parameters exaggerated or with wrong signs.
- Endogeneity:
  - Instrumental variables approach infeasible; residuals regressions used to address endogeneity modestly.
  - Residual-based bid-ask terms jointly highly significant in many specifications, weaker in very highly-rated bond sub-samples.

### Policy implications and suggested extensions
- Practical implication for DMOs:
  - Instrument design (issue size, maturity, jurisdiction, inclusion of enhanced CAC) significantly influences yields and liquidity risk; good debt management can reduce funding costs by tens of basis points, yielding savings of millions of dollars even for a small issuer.
  - Issuers can lower borrowing costs by issuing bonds with features that enhance liquidity at little cost.
- Suggested empirical extensions:
  - Investigate whether bid-ask spreads vary with global and country-specific stress and volatility.
  - Include macroeconomic terms known to market participants at observation dates.
  - Use country-date dummy variables or focus country-by-country for large issuers to isolate issue-specific features.
  - Refine estimates on highly comparable country sub-samples; study corporate Eurobonds, Eurobonds in other currencies, advanced-country Eurobonds, domestic sovereign securities in emerging markets, and investor base composition effects.

*Source: wpiea2022098-print-pdf - 12. Log Yield Determinants; Residuals from the Reduced Form Regression for LYLD_EMD Regressed on Residuals from the Reduced Form Regression for LBAS_EMD and Related Variables*

### References .............................................................................................................

### References

### Boxes
- 1. CACs and Pari-Passu Clauses ..................................................................................................................... 10

### Figures
- 1. Variable Distributions and Scatter Plots ..................................................................................................... 17
- 2. Bid-Ask Spread Determinants; Reduced Form Regressions ................................................................... 24
- 3. Effects of Issue Volume and Country Volume on the Bid-Ask Spread by Issuer Size ........................... 27
- 4. Residual Plots for the Reduced Form Regression on the Log Bid-Ask Spread ..................................... 30
- 5. Residual Plots for the Reduced Form Regression on the Log Bid-Ask Spread ..................................... 30
- 6. Bid-Ask Spread Determinants; Reduced Form Regressions Using Only Explanatory Variables Known 
  at Time of Issue ................................................................................................................................................. 32
- 7. Yield Determinants; Reduced Form Regressions ...................................................................................... 35
- 8. Residual Plots for the Reduced Form Regression on the Log Yield ....................................................... 38
- 9. Residual Plots for the Reduced Form Regression on the Log Yield; Sub-Sample with Rating>14 ...... 38
- 10. Yield Determinants; Reduced Form Regressions Using Only Explanatory Variables Known at Time 
  of Issue ............................................................................................................................................................... 40
- 11. Yield Determinants; Effect on Yield of the Bid-Ask Spread, Controlling for All Other Explanatory 
  Variables ............................................................................................................................................................ 44

### Tables
- 1. Summary of Main Effects ............................................................................................................................. 46
- 2. Credit Ratings and Numerical Codes .......................................................................................................... 48
- 3a. Summary Statistics; Full Sample .............................................................................................................. 49
- 3b. Summary Statistics of Transformed Variables; Main Regression Sample ........................................... 50
- 3c. Summary Statistics of Transformed Variables; Sub-Sample with Credit Rating > 14.......................... 51
- 3d. Summary Statistics of Transformed Variables; Sub-Sample with Total Country Volume ≤ US$10 
  billion .................................................................................................................................................................. 52
- 4: Log Bid-Ask Spread Determinants; Reduced Form Regressions ............................................................ 53
- 5. Log Bid-Ask Spread Reduced Form Regressions F Test Results, Full Sample ..................................... 55
- 6: Log Bid-Ask Spread Determinants; Reduced Form Regressions Using only Explanatory Variables 
  Known at Time of Issue .................................................................................................................................... 56
- 7. Log Bid-Ask Spread Reduced Form Regressions Using only Explanatory Variables Known at Time of 
  Issue; F Test Results; Full Sample Regression ............................................................................................. 57
- 8: Log Yield Determinants; Reduced Form Regressions .............................................................................. 58
- 9. Log Yield Reduced Form Regressions; F Test Results, Full Sample ...................................................... 60
- 10. Log Yield Determinants; Reduced Form Regressions Using only Explanatory Variables Known at 
  Time of Issue ..................................................................................................................................................... 61
- 11. Log Yield Reduced Form Regressions Using only Explanatory Variables Known at Time of Issue; F 
  Test Results ....................................................................................................................................................... 62

*IMF WORKING PAPERS — Sovereign Eurobond Liquidity and Yields*

### 12. Log Yield Determinants; Residuals from the Reduced Form Regression for LYLD_EMD Regressed

### 12. Log Yield Determinants; Residuals from the Reduced Form Regression for LYLD_EMD Regressed on Residuals from the Reduced Form Regression for LBAS_EMD and Related Variables

### Introduction and context
- Eurobonds provide international diversification, relative liquidity, transparency, and flexible access to a wide investor base for emerging market and developing country (EMDC) sovereigns.
- Characteristics of Eurobonds:
  - Issued in a jurisdiction (typically New York or England) and a currency (predominantly the US dollar) distinct from those of the issuer.
  - Sold “over the counter” with transactions normally recorded in major international central securities depositories.
- Market scale and composition (as reported):
  - By 2020 there were about US$1.5 trillion in rated sovereign EMDC Eurobonds outstanding—up from under US$1 trillion in 2015.
  - About 70 new issues per year over the past decade.
  - As of 2017, 82 countries had issued Eurobonds.
  - About US$500 billion in non-sovereign Eurobonds were outstanding (issued by major EMDC banks and enterprises).
- Importance:
  - Eurobond pricing is monitored as an indicator of market perception of risk and used as a benchmark for other products.
  - Eurobond markets’ homogeneity in some dimensions and heterogeneity in others make them useful for studying financial market functioning.

### Liquidity concept and connection to yields
- Market liquidity definition: ability to buy or sell a reasonable amount of a security, reasonably quickly, with minimal effect on price.
- Bid-ask spread is a necessary but not sufficient metric for liquidity; what is “modest” varies by market (a few basis points in advanced wholesale bond markets; some tens of basis points in emerging market bond markets).
- Liquidity value to market participants:
  - Facilitates price discovery and enables investors to adjust positions quickly.
  - Eases debt management operations by DMOs (selling more securities; liability management operations).
  - Critical during crisis periods: liquidity disappearance can amplify shocks and contribute to market freeze or systemic collapse (references: Brunnermeier and Pedersen, 2009; Moutot et al., 2017; Rösch and Kaserer, 2013).
- Hypothesis: bond characteristics that promote liquidity (lower bid-ask spreads) should be rewarded by lower yields.

### Theoretical determinants of bid-ask spreads and liquidity
- Reasons dealers charge bid-ask spreads:
  - Operational costs (some fixed element), implying spreads inversely related to turnover.
  - Market power of dealers (limits ability of counterparties to shop around).
  - Financing and capital costs for holding inventory; riskiness of security increases spread.
  - Adverse selection risk (compensate for trading with better-informed counterparties).
- Implications:
  - Bid-ask spread should relate to turnover, investor variety, issuance volume, credit and market risk, duration, and standardization of features.
  - Feedback loops: liquidity lowers yields, but attractive returns can increase demand and thus improve liquidity.

### Empirical evidence — domestic bond markets
- General findings:
  - Yields tend to increase with illiquidity across assets and over time.
  - Both common (market-wide) and issuer-specific liquidity factors matter; the common factor is more important in stress times (Heck et al., 2015).
  - Liquidity pricing is especially important for speculative bonds and in periods of crisis (De Jong and Driessen, 2006; Friewald et al., 2012).
- Specific observed relationships in domestic markets:
  - Amihud and Mendelson: yield negatively related to liquidity for U.S. government securities; excess yield on less liquid notes decreases with time to maturity.
  - “Seasoning” (age of bond) increases bid-ask spread; increasing maturity raises spreads (Bildersee, Sanig and Warga, Devani and Zhang).
  - Recently issued (“on-the-run”) Treasuries have lower returns (greater liquidity) than “off-the-run” issues (Warga, 1992).
  - Bid-ask spread commonly highly correlated with other price-based liquidity measures; end-of-day bid-ask spread performs robustly (Lesmond, Chen et al., Fleming, Langedijk et al., 2018).
  - Determinants across studies include: turnover (negative effect on spread), credit rating (negative effect on spread), bond age (mixed), maturity and duration (positive effect on spread), coupon rate (effects through duration).

### Empirical evidence — Eurobond markets and international studies
- Typical Eurobond pricing literature focuses on risk premia relative to a safe asset (e.g., U.S. government bond), reflecting default probability, expected loss given default, risk compensation, and global macro-financial conditions.
- Studies linking liquidity to pricing in international contexts:
  - Historical and emerging market evidence shows higher bid-ask spreads associated with higher yields; small issue size also associated with higher yields (Alquist, 2008; Duffie et al., 2003).
  - Hund and Lesmond (2008): liquidity is statistically and economically significant in explaining differences in yield spreads for sovereign and corporate emerging market bonds; bid-ask spread correlates well with other liquidity measures and remains significant after joint estimation and instrumental approaches.
  - Ap Gwilym et al. (2002): spread negatively related to credit rating and issue size; positively related to price volatility.
  - Chamon et al. (2018): issuance in a foreign jurisdiction affects yields; higher bid-ask spreads on foreign-jurisdiction bonds make them less attractive, especially during stress times.
- Term-and-condition features:
  - Inclusion of collective action clauses (CACs) tends to produce a small but significant reduction in yields, particularly for lower-rated countries and during stress periods (Chung and Papaioannou, 2020; Becker et al., 2003; Richards and Gugiatti, 2003; Bardozzetti and Dottori, 2014; Carletti et al., 2018; Picarelli et al., 2018).
  - Some evidence that CACs may reduce borrowing costs for more creditworthy issuers but increase them for less creditworthy borrowers (Eichengreen and Mody, 2004, based on earlier data).
  - Jurisdiction choice matters and can be “as important as” CAC inclusion; initial yields on bonds issued under New York law may be substantially lower than those under English law (Ratha et al., 2016; Tsatsaronis, 1999).

### Implications for issuers, DMOs, and market participants
- Characteristics and practices that likely lower yields by enhancing liquidity:
  - Larger issuance size (increases free float and turnover).
  - Standardized features and commonly used jurisdictions/currencies (promote homogeneity).
  - Shorter seasoning or issuance of “on-the-run” issues to maintain liquidity.
  - Inclusion of contract terms valued by investors (e.g., CACs under certain conditions) depending on issuer creditworthiness and market perception.
- Importance for policy and debt management:
  - DMOs can potentially lower borrowing costs by issuing bonds with features that enhance liquidity at little cost to the issuer.
  - Monitoring and understanding bid-ask spreads provides actionable insight into market functioning and expected yield premia.
  - During stress periods, liquidity effects on yields amplify, so debt management strategies should account for changing liquidity premia.

*Source: wpiea2022098-print-pdf - 12. Log Yield Determinants; Residuals from the Reduced Form Regression for LYLD_EMD Regressed on Residuals from the Reduced Form Regression for LBAS_EMD and Related Variables*

### Box 1. CACs and Pari-Passu Clauses

### Box 1. CACs and Pari-Passu Clauses

### Purpose and rationale
- Eurobond terms and conditions contain clauses meant to facilitate restructuring when that is necessary, while preserving appropriate incentives for the issuer and all bond holders.
- These clauses can reduce value-destruction in restructuring situations (caused by the disruption to financing and payments), and therefore should be welfare enhancing.
- Clauses are designed to ensure an equitable distribution of the benefits, and discourage perverse behavior such as “holding out” for an exceptional good deal by a small group of investors, or premature calling for restructuring by a debtor.
- These clauses are especially important for sovereign bonds due to the absence of an insolvency procedure that would organize creditor claims.

### Collective Action Clauses (CACs)
- A CAC enables a qualified majority of bondholders to bind the minority to the terms of a restructuring, making it more difficult to block a restructuring.
- Original CACs operated on a series-by-series basis, allowing a group of creditors to relatively easily obtain a “blocking position,” which could discourage more cooperation-minded creditors from agreeing to terms that risk being inequitable.
- Enhanced CACs allow for:
  - “two limb” aggregation (requiring agreement by series and in aggregate); or
  - “single limb” aggregation (requiring just a supermajority of the aggregate holders of all bonds).
- Use of a single limb voting procedure requires the issuer to offer all affected bondholders the same instrument or an identical menu of instruments (conditions that are “uniformly applicable”).
- Empirical evidence: the inclusion of enhanced CACs does indeed seem to reduce holdout risk (Fang et al., 2021).

### Pari-passu clauses
- A pari-passu clause, as applied to a sovereign bond, protects a creditor from legal subordination of its claims in favor of another creditor.
- Effectively, the clause impedes the debtor from doing independent deals with different creditor groups, increasing creditor cohesion and strengthening the bargaining powers of creditors.
- A “modified” pari-passu clause makes explicit that payments are not “ratable.” Thus, to the extent that a debt is due and payable to one creditor, the clause allows a debtor to make payments to that creditor even if it does not make pro rata payments to other creditors whose debts are also due and payable.
- The modified clause therefore allows the debtor to service restructured bonds even as it does not make payments on bonds held by “holdouts.”

### Jurisdictional and contractual variation
- The exact formulation and implications of CACs and pari-passu clauses differ slightly across jurisdiction.
- Legislation, precedent, and practice differ across jurisdictions in other ways that may affect sovereign debt restructuring.
- Example: bonds issued under New York law more often use a trust structure than do bonds issued under English law, and therefore typically may incorporate certain protections against holdout creditors.

### History, practice, and consensus
- Use of CACs and pari-passu clauses in sovereign debt contracts has a long history.
- CACs were promoted by the IMF and others following the wave sovereign debt crises in the late 1980s and 1990s, to counter a tendency for debt restructurings to be excessively costly.
- The inclusion of CACs in their original form became common practice after 2003, also under New York law.
- Consultation between the official and the private sectors in response to creditor coordination problems in the Argentine and the Greek debt restructurings led to nearly all Eurobonds issued since 2015 including enhanced CACs and modified pari-passu clauses (IMF, 2019).
- The current contractual clauses and their widespread use reflect a consensus on best practice forged between issuers, investors, intermediaries, international financial institutions, and non-government organizations.

*International Monetary Fund*

### 18. A slight uptick can be seen in the spread for top-rated issues. Possibly, the typical investor for

### 18. A slight uptick can be seen in the spread for top-rated issues. Possibly, the typical investor for

### Determinants of bid-ask spreads
- Initial and current credit ratings both have economically significant effects; the change in the rating since issuance has a strong effect (top right-hand chart on the second page of Figure 2).
- The initial rating matters for countries with below-average credit ratings; once that mid-point is reached, the effect is minimal.
- The relationship has little convexity: the initial rating matters most for low-rated issuers, but the change affects all bonds roughly equally.
- Coupon rate:
  - Distinctly positive but nonlinear relationship with the bid-ask spread (bottom row right-hand chart).
  - A coupon rate of 1.5 percent is associated with a bid-ask spread about 25 basis points lower than in the case of a coupon rate of 5 percent.
  - The coupon rate on a Eurobond may signal creditworthiness beyond the credit rating.
- Inclusion in the EMBIG:
  - Small effect on the bid-ask spread, positive for low-rated issuers and negative for medium- to high-rated issuers (middle left-hand chart on the second page of Figure 2).
  - Estimated parameters are not individually significantly different from zero, but jointly significant (Table 5).
- Issuance jurisdiction:
  - Issuance under New York is found to increase the bid-ask spread for low-rated bonds.
  - No significant effect on bid-ask spreads of medium- to high-rated bonds.
- Contractual clauses:
  - Inclusion of an original CAC has only a slight negative effect; estimated parameters are insignificant.
  - Inclusion of an enhanced CAC is highly significant and narrows bid-ask spreads, especially for low-rated bonds:
    - In case of very low rating, the bid-ask spread can be reduced by about 20 basis points, compared to a sample average spread of 55 basis points.
    - Spreads on medium-rated bonds are reduced by about 5 basis points.

### Vintages, maturity, volume and other characteristics
- Issue year effects:
  - Bonds issued in 1999-2003 seem to have unusually large bid-ask spreads, but 1998 seems to have been a good year to issue.
  - Estimated parameters from 2004 onward are stable, even during the global financial crisis.
- Issue volume:
  - Increasing issue volume has an overall negative effect on spreads, reduced to about 20 basis points when country dummies are included, independent of country rating.
  - Deviation from the country average issue volume tends to increase spreads, now also for lowest-rated countries (with country dummies).
- Remaining maturity and age:
  - Quadratic / nonlinear relationships; individual parameter estimates on remaining maturity and change in maturity (age) are relatively unstable due to multicollinearity, but overall effects are similar.

### Robustness and diagnostics for bid-ask spread regressions
- Robust regression estimates are mostly close to OLS estimates.
- Results robust to splitting sample into early/late observation periods and to including only bonds issued after 2009.
- Trimming or extending tail observations affects log-linear and log-quadratic term estimates as expected.
- Inclusion of country dummies:
  - Increases the R^2 by about 10 percentage points, to 0.77.
  - Reduces magnitude and significance of many individual parameters while preserving overall signs of effects.
- Residuals:
  - Residuals are very close to normal, possibly with some heteroskedasticity.
  - Removing the absolute largest 12 residuals (about 0.5 percent of the sample) addresses leptokurtosis.
  - For the sub-sample with rating>14 (above BBB), residuals show very asymmetric and very large deviations from normality (Figure 5).

### Predicting bid-ask spreads using only variables known at time of issue
- Predictive performance:
  - R^2 for the full sample is just over 0.5—about 18 percentage points lower than when all explanatory variables are used.
  - Many estimated parameters are significant individually or jointly (Table 7).
- Issue volume:
  - Major influence for issues of less than US$1 billion.
  - Threshold effects are economically and statistically significant at US$500 million and also at higher levels.
- Total country volume and deviation from average issue size:
  - Coefficients significant; overall effect modest for full sample.
  - Splitting sample: spreads narrow with total volume for small issuers, modest and slightly positive effect for large issuers.
  - Deviation-from-average relationship is preserved and more pronounced than with full explanatory set.
- Remaining time to maturity and seasoning:
  - Bid-ask spread remains a quadratic function of remaining time to maturity.
  - Seasoning has pronounced, non-linear effect: liquidity predicted to dry up in final few years before maturity.
- Credit rating and coupon:
  - High credit rating or low coupon rate helps reduce predicted bid-ask spread.
  - Coupon rate effect tapers off sharply for rates above about 3 percentage points.
- Predictable contractual effects:
  - Inclusion in EMBIG, issuance under New York law, inclusion of original or enhanced CAC are predictable at issuance.
  - Issuing under New York law widens bid-ask spread for low-rated bonds, but enhanced CAC can more than offset this increase.
- Robustness:
  - Results mostly robust across variations; robust regression technique often yields larger absolute parameter estimates.
  - High-rating bonds exhibit multicollinearity and less reliable residual distributions.

### Determinants of yields (using all explanatory variables)
- Good explanatory power: regressions for log yields achieve R^2 around 0.85.
- Issue volume:
  - Important determinant, especially for middle-range ratings and volumes below US$1 billion.
  - Reducing volume from US$1.5 billion to US$750 billion raises the yield by over 60 basis points for a mid-rating bond.
  - Threshold effects most pronounced at US$500 million; small penalty for very large issue sizes.
  - Issue volume not important for very high ratings; yields on lowest-rated rise with volume.
- Remaining volume outstanding:
  - Modest effect until remaining volume falls below a third or a quarter of initial amount, then rates rise steeply.
- Total country volume:
  - For most countries yields tend to increase with total amount outstanding, but effect not very large and coefficients mostly insignificant.
  - Splitting sample: positive relationship maintained except for smaller, lower-rated issuers where aggregate volume increases can markedly decrease yields.
- Deviation from country average issue size:
  - Small issue relative to country average can achieve somewhat lower yield; effect modest for highly rated bonds.
  - Very low-rated bonds: large issue may signal desperate financing need.
- U.S. term structure:
  - Effect of current U.S. bond yields is very pronounced and independent of credit rating.
- Remaining maturity and seasoning:
  - Remaining maturity affects yield monotonically and largely independent of credit rating.
  - Seasoning non-monotonic: for a 15-year initial term, yield falls for first decade then rises sharply as maturity approaches.
- Credit rating and coupon:
  - Lower credit rating associated with higher yield.
  - Current rating matters most; F tests indicate joint insignificance of when-issued rating variables in yields regression, but cross-product effects may matter.
  - Low coupon rate associated with low yields; curvature sharp above about 4.5 percent for most rating levels.
- Contractual clauses and jurisdiction:
  - Inclusion in EMBIG has significant but modest overall effect on yields.
  - Issuance under New York law raises yields markedly, especially for low-rated bonds (possibly to compensate for higher bid-ask spreads or higher expected restructuring costs under New York law).
  - Original CAC has small negative effect on yields (jointly significant); enhanced CAC markedly reduces yields, especially on low-rated bonds—depending on rating, yields can be reduced by tens of basis points.
- Issue year effects:
  - Good and bad vintages largely coincide with those for bid-ask spreads; 1999-2003 show both large spreads and high yields.
- Robustness:
  - Results qualitatively robust across sample variations.
  - Including country dummies increases R^2 by about 7 percentage points; most retained parameter signs little affected.
  - Higher issue volume associated with lower yields, maximum effect about 110 bps.
  - Contractual-term parameters can be affected by collinearity with country dummies.
  - Regression for rating>BBB (=14) explains high proportion of variance but many parameters seem exaggerated or have wrong signs.
- Residuals:
  - Distribution close to symmetric with modest leptokurtosis; about 0.5 percent of residuals lie far from normal.
  - For sub-sample rating>14 residuals are highly non-normal and asymmetric (Figure 9).

### Predicting yields using only variables known at time of issue
- Predictive performance:
  - Excluding contemporaneous and fixed effects reduces R^2 on LYLD_EMD to around 0.6.
  - Qualitative relationships to explanatory variables largely preserved (Tables 10 and 11, Figure 10).
- Issue volume and thresholds:
  - Initial issue volume remains major predictor, especially for smaller issues.
  - Peak effects at low issuance volumes and threshold effects at US$500 million are larger when using only issuance-time variables.
- Country volume and deviation from average:
  - Slopes of projected effects are positive; individual estimates mostly similar to full specification.
  - Country volume terms mostly insignificant; for smaller issuers a negative relationship between yields and country volume is again found.
- Maturity, seasoning, coupon, rating:
  - Residual maturity quadratic relationship to projected yield.
  - Yield rise as bond approaches maturity and liquidity dries up can be well-predicted.
  - Higher coupon rate associated with persistently higher yields.
  - Credit rating projected effect as expected, but individual parameters not precisely estimated.
- Contractual clauses and jurisdiction:
  - Inclusion in EMBIG is not a reliable indicator of future yields.
  - Issuance under New York law and inclusion of enhanced CAC remain significant and similar to full specification.
- Robustness:
  - Results not greatly affected by robust estimation, restricting to bonds issued after 2009, or to credit rating below 15.
  - Estimates for higher rated bonds (about one fifth of sample) are less reliable.

### Interaction of bid-ask spreads and yields
- Objective: investigate relationship between yields and market liquidity (bid-ask spread) after controlling for all available explanatory variables.
- Identification challenges:
  - Instrumental variables approach not feasible: candidate instruments fail over-identification tests or lack economic rationale.
  - Thus a more modest approach: regress yield residuals on bid-ask-spread-related residuals after orthogonalizing each dependent variable to all explanatory variables.
- Procedure:
  - Regressions run for log yield; log bid-ask spread; square of log bid-ask spread; cross-product of log bid-ask spread and log initial rating; cross-product of log bid-ask spread and squared log initial rating on all explanatory variables.
  - Residuals (orthogonal to explanatory variables) computed and used to regress yield residual on bid-ask spread-related residuals.
- Main quantitative result:
  - An increase (decrease) in the bid-ask spread of 10 basis points, starting from the sample mean of 55 basis points, increases (decreases) the yield by about 20 basis points for a typical bond with a mid-range credit rating.
- Statistical significance and robustness:
  - F tests show parameter estimates for groups of variables are highly significant even if some individual estimates are insignificant.
  - Parameters involving cross-products with credit rating are jointly significant in several sub-samples but do not strongly affect the overall relationship.
  - Results fairly robust across sample variations except when restricted to very highly-rated bonds.
- Comparable findings in literature:
  - Hund and Lesmond: a 1 bp increase in the bid-ask spread raises corporate bond yield spreads by 3.13 bps for corporate bonds and 5.67 bps for government bonds, after controls; changes in bid-ask spreads account for over 10 percent of total variation in yield spreads.
  - Chen et al: a 1 bp increase in the bid-ask spread of US corporate bonds increases yield spreads by 0.42 bps for investment grade bonds.

*International Monetary Fund*

### 2.30 bps. for speculative grade bonds. When they allow for the endogeneity of the bid-ask spread, the

### wpiea2022098-print-pdf - 2.30 bps. for speculative grade bonds. When they allow for the endogeneity of the bid-ask spread, the

### Findings on yields, bid-ask spreads, and instrument design
- Instrument design significantly influences yields and liquidity risk (bid-ask spreads) for US$ EMDC eurobonds. Important design choices: issue size; maturity; jurisdiction of issuance; inclusion of an enhanced CAC.
- Allowing for endogeneity of the bid-ask spread reduces the coefficient on spreads to about 0.2 for both bond categories; the estimated range of effects, from most to least liquid bonds, is about 160 basis points.
- Yields reflect duration, credit risk, and liquidity risk; characteristics of very highly rated bonds are less reliably related to yields and spreads.
- The choices at issuance have persistent effects: yields and spreads “store information about history.”
- Good debt management at issuance and through subsequent intervention can reduce funding costs by tens of basis points, yielding savings of millions of dollars even for a small issuer.

### Main effect magnitudes (summary of Table 1)
- Decrease BAS by 10 bps. requires / Decrease YLD by 30 bps. requires (contextual framing in table).
- Issue volume effects (dummy step effects on BAS_EMD and YLD_EMD):
  - Issue volume ≥US$0.5 bn.: Effect on BAS_EMD = -9.0 bps.; Effect on YLD_EMD = -70 bps.
  - Issue volume ≥US$1.0 bn.: Effect on BAS_EMD = -5.0 bps.; Effect on YLD_EMD = -18 bps.
  - Issue volume ≥US$1.5 bn.: Effect on BAS_EMD = -3.0 bps.; Effect on YLD_EMD = -23 bps.
  - Issue volume ≥US$3.0 bn.: Effect on BAS_EMD = -0.5 bps.; Effect on YLD_EMD = +13 bps.
- Inclusion in EMBI Global: Effect on BAS_EMD = -1.5 bps. when CRR=BB+; +3.0 bps. when CRR= B-; Effect on YLD_EMD = -1.5 bps. when CRR= BB+; +0.5 bps. when CRR= B-.
- Issued under NY law: Effect on BAS_EMD = +2.0 bps. at CRR= BB+; +11.0 bps. when CRR= B-; Effect on YLD_EMD = +17.0 bps. at CRR= BB+; +67.0 bps. when CRR=B-.
- Inclusion of an original CAC: Effect on BAS_EMD = -1.5 bps. at CRR= BB+; -3.0 bps. when CRR= B-; Effect on YLD_EMD = -19.0 bps. at CRR= BB+; +6.5 bps. when CRR= B-.
- Inclusion of an enhanced CAC: Effect on BAS_EMD = -5.5 bps. at CRR= BB+; -18.0 bps. when CRR= B-; Effect on YLD_EMD = -19.5 bps. at CRR= BB+; -61.0 bps. when CRR= B-.
- Credit rating effects:
  - One rating grade improvement reduces yield approximately 60 bps. (note 2/).
  - One rating grade improvement reduces yield approximately 45 bps. (note 3/).

### Regression evidence — Log Bid-Ask Spread determinants (Table 4 highlights)
- Sample size for main regression: 4,363 observations; Adjusted R2 reported (varies by specification).
- Key estimated coefficients (full sample, reduced form):
  - LVOL_ISS: coefficient examples include 6.578*** (s.e. 1.473) in one specification; positive and statistically significant in many specifications.
  - LVOL_ISS_LCRR_I: -5.106*** (s.e. 1.153) in one specification; significant negative interaction with credit rating index in many specifications.
  - CHG_LVOL: -25.988*** (s.e. 3.892) in one specification; strongly negative.
  - D_E_CAC: -5.131*** (s.e. 0.872) with D_E_CAC_LCRR_I = 4.111*** (s.e. 0.709); strong nonlinear interactions with credit rating.
  - LMAT_NOW_SQ: -0.060*** (s.e. 0.013) — nonlinear maturity effect.
  - D_NY_LAW: 1.679*** (s.e. 0.582) with D_NY_LAW_LCRR_I = -1.268*** (s.e. 0.474) and quadratic interaction D_NY_LAW_LCRR_I_SQ = 0.226** (s.e. 0.095).
- Robustness and sub-samples: coefficients and significance patterns reported for extended sample, first half, second half, trimmed, and robust regressions. Adjusted R2 range examples: 0.610 to 0.723 across specifications.

### Regression evidence — Log Yield determinants (Table 8 highlights)
- Full sample: 4,363 observations; Adjusted R2 often very high (e.g., 0.873 in several specifications).
- Key estimated coefficients (full sample, reduced form):
  - LVOL_ISS: 4.440*** (s.e. 0.620) — larger issue volumes associated with lower yields via interactions.
  - LVOL_ISS_LCRR_I: -3.463*** (s.e. 0.488) and LVOL_ISS_LCRR_I_SQ: 0.673*** (s.e. 0.095).
  - CHG_LVOL: -11.578*** (s.e. 2.129) with CHG_LVOL_LCRR_I = 8.572*** (s.e. 1.656) and CHG_LVOL_LCRR_I_SQ = -1.612*** (s.e. 0.320).
  - D_E_CAC: -2.298*** (s.e. 0.496) with D_E_CAC_LCRR_I = 1.924*** (s.e. 0.401) and D_E_CAC_LCRR_I_SQ = -0.394*** (s.e. 0.080).
  - D_NY_LAW: 1.989*** (s.e. 0.310) with D_NY_LAW_LCRR_I = -1.650*** (s.e. 0.252) and D_NY_LAW_LCRR_I_SQ = 0.332*** (s.e. 0.051).
- Nonlinearities and interactions with country credit rating (LCRR_ISS and changes) are prominent.
- Robustness across sub-samples (rating >14, rating <15, issued after 2009, small/large issuers) reported with varying coefficient magnitudes and significance; Adjusted R2 remains high across many specifications.

### Endogeneity and residuals regressions (Table 12 and discussion)
- Neither spreads nor yields are exogenous; residuals regressions run to address endogeneity (LYLD_EMD residuals regressed on LBAS_EMD residuals and related variables).
- Results: R_LBAS and R_LBAS_SQ often statistically significant in residuals regressions:
  - Example coefficient: R_LBAS = 0.182*** (s.e. 0.011) in one reported column.
  - R_LBAS_SQ = 0.101*** (s.e. 0.035) in full sample column.
- F test results for joint significance of residual-based bid-ask terms:
  - Full sample F(4, 4358) = 80.73*** (Adjusted R2 = 0.100 in that specification).
  - Sub-sample with rating >14: F(4, 3380) = 75.62*** (Adjusted R2 = 0.135).
  - Issued after 2009 and other sub-samples show significant joint F tests in many cases; some small-sample categories show weaker F test significance.

### Summary of key data and descriptive statistics (selected from Tables 3a–3d)
- Full sample (Table 3a; No. Obs. = 4,759):
  - BAS_EMD mean = 62.78; Std. Dev. = 47.49; Min = 0.10; Max = 1,329.
  - YLD_EMD mean = 531.22; Std. Dev. = 506.74; Min = 17.40; Max = 21,831.
  - VOL_ISS mean = 1.62; Std. Dev. = 1.08; Min = 0.10; Max = 7.00.
  - CRR_ISS mean = 12.30; Std. Dev. = 3.45; Min = 6; Max = 20.
  - D_EMBIG mean = 0.54; D_NY_LAW mean = 0.55; D_O_CAC mean = 0.50; D_E_CAC mean = 0.36.
- Main regression sample transformed variables (Table 3b; No. Obs. = 4,363):
  - LBAS_EMD mean = 4.018; Std. Dev. = 0.371; Min = 3.220; Max = 5.162.
  - LYLD_EMD mean = 6.156; Std. Dev. = 0.341; Min = 4.867; Max = 7.019.
  - LVOL_ISS mean = 0.279; Std. Dev. = 0.649; Min = -2.303; Max = 1.946.
  - LCRR_ISS mean = 2.474; Std. Dev. = 0.275; Min = 1.792; Max = 2.996.
  - D_E_CAC mean = 0.383; D_NY_LAW mean = 0.547; D_O_CAC mean = 0.489; D_EMBIG mean = 0.551.
- Sub-sample with rating > 14 (Table 3c; No. Obs. = 978):
  - LBAS_EMD mean = 3.904; LYLD_EMD mean = 5.944.
  - D_E_CAC mean = 0.442; D_O_CAC mean = 0.443.
- Sub-sample with total country volume ≤ US$10 billion (Table 3d; No. Obs. = 1,222):
  - LBAS_EMD mean = 4.218; LYLD_EMD mean = 6.256.
  - D_E_CAC mean = 0.373; D_O_CAC mean = 0.514.

### Statistical tests and joint significance (selected F test results)
- Log Bid-Ask Spread reduced form, full sample (Table 5):
  - LVOL_ISS block: F(4, 4278) = 23.49; Probability>F = 0.0000.
  - CHG_LVOL block: F(4, 4278) = 39.89; Probability>F = 0.0000.
  - LCRR_ISS block: F(4, 4278) = 97.35; Probability>F = 0.0000.
  - D_E_CAC block: F(3, 4278) = 15.55; Probability>F = 0.0000.
- Log Yield reduced form, full sample (Table 9):
  - LVOL_ISS block: F(4, 4278) = 14.42; Probability>F = 0.0000.
  - CHG_LVOL block: F(4, 4278) = 30.63; Probability>F = 0.0000.
  - LMAT_NOW_DYC block: F(4, 4278) = 142.20; Probability>F = 0.0000.
  - LCRR_ISS and CHG_LCRR block: F(4, 4278) = 553.88; Probability>F = 0.0000.
  - D_E_CAC block: F(3, 4278) = 9.26; Probability>F = 0.0006.

### Policy implications and suggested extensions
- Practical implication: debt managers (DMOs) can use instrument design to influence funding costs and market liquidity; small issuers can achieve savings of millions of dollars by reducing yields and spreads through issuance choices.
- Suggested empirical extensions:
  - Investigate whether bid-ask spreads vary with global and country-specific stress and volatility (risk premia, macro-financial indicators).
  - Include macroeconomic terms as explanatory variables, ensuring they are known to market participants at observation dates.
  - Use country-date dummy variables or focus country-by-country on large issuers to isolate issue-specific features.
  - Refine estimates on highly comparable country sub-samples for investor or DMO research.
  - Investigate short-term price movements with a different dataset to control for slow-moving global/country effects.
  - Study determinants of bid-ask spreads and yields for corporate Eurobonds; Eurobonds in other currencies; advanced-country Eurobonds; domestic sovereign securities in emerging markets; and the relationship between investor base composition/size and yield/liquidity.

*Source: wpiea2022098-print-pdf (excerpt provided).*

### References

### References

### Works cited

- Afonso, Antonio, Michael G. Arghyrou, and Alexandros Kontonikas (2015), “The Determinants of Sovereign Bond Yield Spreads in the EMU,” European Central Bank, Eurosystem, Working Paper Series.  
- Alquist, Ron (2008), “How Important is Liquidity Risk for Sovereign Bond Risk Premia? Evidence from the London Stock Exchange,” Bank of Canada Staff Working Paper 2008–47. 
- Amihud, Yakov, and Haim Mendelson (1980), “Dealership market: Market-making with inventory,” Journal of Financial Economics, Vol. 8.1, pp. 31-53. 
- Amihud, Yakov, and Haim Mendelson (1986a), “Asset pricing and the bid-ask spread,” Journal of Financial Economics, Vol. 17, pp. 223-249. 
- Amihud, Yakov, and Haim Mendelson (1986b), The Effects of Beta, Bid-Ask Spread, Residual Risk, and Size on Stock Returns,” The Journal of Finance, Vol. 44.2, pp. 479-486. 
- Amihud, Yakov, and Haim Mendelson, (1991), “Liquidity, Maturity, and the Yields on U.S. Treasury Securities,” Journal of Finance, Vol. 46.4, pp. 1411-425. 
- Amihud, Yakov, and Haim Mendelson (2015), “The Pricing of Illiquidity as a Characteristic and as Risk,” Multinational Finance Journal, Vol. 19.3, pp. 149-168. 
- Amihud, Yakov, Allaudeen Hameed, Wenjin Kang, and Huiping Zhang (2015), “The Illiquidity Premium: International evidence,” Journal of Financial Economics, Vol. 117(2), pp. 350-368. 
- Ammer, John, and Fang Cai (2007), “Sovereign CDS and Bond Pricing Dynamics in Emerging Markets: Does the Cheapest-to-Deliver Option Matter?” Board of Governors of the Federal Reserve System, International Finance Discussion Papers No. 912. 
- Ap Gwilym, Owain, Lourdes Trevino, and Stephen H. Thomas (2002) “Bid-Ask Spreads and the Liquidity of International Bonds,” The Journal of Fixed Income (September), pp. 82-91. 
- Bao, Jack, Jun Pan, and Jiang Wang (2011), “The Illiquidity of Corporate Bonds,” The Journal of Finance, Vol. 66.3, pp. 911-46. 
- Bardozzetti, A., and D. Dottori (2014), “Collective action clauses: How do they affect sovereign bond yields?” Journal of International Economics, Vol. 92(2), pp. 286–303. 
- Becker, Torbjorn I., Anthony J. Richards, and Yunyong Thaicharoen (2001), “Bond Restructuring and Moral Hazard: Are Collective Action Clauses Costly?” IMF Working Paper WP/01/92, August. 
- Bekaert, Geert, Campbell Harvey, and Christian Lundblad (2007), “Liquidity and Expected Returns: Lessons from Emerging Markets,” The Review of Financial Studies, Vol. 20.6, pp. 1783-1831. 
- Becker, T., A. Richards, and Y. Thaicharoen (2003), “Bond restructuring and moral hazard: are collective action clauses costly?” Journal of International Economics, Vol. 61.1, pp. 127–161. 
- Benston, George J., and Robert L. Hagerman (1978), “Risk, Volume and Spread,” Financial Analysts Journal, Vol. 34.1, pp. 46-49. 
- Bildersee, John S. (1977), “Price Spreads, Performance, and the Seasoning of New Treasury and Agency Bond Issues,” Journal of Financial and Quantitative Analysis, Vol. 12.3, pp. 433-55. 
- Bildersee, John S. (1979), “Bid-Ask Price Spreads in the Agency Bond Market,” Journal of Money, Credit & Banking, Vol. 11, pp. 209-213. 
- Bildersee, John S. (1980), “Price Elasticity and Bid-Ask Price Spreads in the Treasury Bond Market,” Journal of Business Research, Vol. 8.1, pp.105–111. 
- Bolton, Paul, and Olivier Jeanne (2007), “Structuring and Restructuring Sovereign Debt: The Role of a Bankruptcy Regime,” Journal of Political Economy, Vol. 115.6, pp. 901–924.  
- Bongaerts, Dion, Frank De Jong, and Joost Driessen (2011), “Derivative Pricing with Liquidity Risk: Theory and Evidence from the Credit Default Swap Market,” Journal of Finance, Vol. 66.1, pp. 203-40. 
- Brandt, Michael W., and Kenneth A. Kavajecz (2004), “Price Discovery in the U.S. Treasury Market: The Impact of Orderflow and Liquidity on the Yield Curve,” Journal of Finance, Vol. 59.6, pp. 2623-654. 
- Brunnermeier, Markus, and L. H. Pedersen (2009), “Market Liquidity and Funding Liquidity,” Review of Financial Studies, Vol. 22.6, pp. 2201–2238. 
- Calomiris, Charles W., Mauricio Larrain, Sergio L. Schmukler, and Tomás Williams (2019), “Search for Yield in Large International Corporate Bonds: Investor Behavior and Firm Responses,” Cambridge, Mass: National Bureau of Economic Research Working Paper No.25979.  
- Carletti, Elena, Paolo Colla, Gaurang Mitu Gulati, and Steven R. G. Ongena (2017), “The Price of Law: The Case of the Eurozone Collective Action Clauses,” SSRN Electronic Journal. 
- Chacko, George (2005), “Liquidity Risk in the Corporate Bond Markets,” SSRN Electronic Journal. 
- Chacko, George, Jakub Jurek, and Erik Stafford (2008), “The Price of Immediacy,” The Journal of Finance, Vol. 63.3, pp. 1253-1290. 
- Chakravarty, Sugato, and Asani Sarkar (2003), “Trading Costs in Three U.S. Bond Markets,” Journal of Fixed Income, Vol. 13.1, pp. 39-48.) 
- Chamon, M., J. Schumacher, and C. Trebesch (2018), “Foreign-law bonds: Can they reduce sovereign borrowing costs?” Journal of International Economics, Vol.114, pp. 164-179. 
- Chen, Long, David A. Lesmond, and Jason Wei (2007), “Corporate Yield Spreads and Bond Liquidity,” The Journal of Finance, Vol. 62.1, pp. 119-149. 
- Choudhry, M. (2008), “The Eurobond Market,” in Fabozzi, Frank J. (Ed.) Handbook of Finance. Hoboken, N.J.: Wiley. 
- Chung, Kay, and Michael Papaioannou (2020), “Do Enhanced Collective Action Clauses Affect Sovereign Borrowing Costs?” IMF Working Paper WP/20/162. 
- Csonto, Balazs, and Ixryna Ivaschenko (2013), “Local Fundamentals and Global Factors vs. Ever-Changing Misalignments,” Washington, D.C.: International Monetary Fund, WP/13/164. 
- De Jong, Frank, and Joost Driessen (2006), “Liquidity Risk Premia in Corporate Bond Markets,” SSRN Electronic Journal. 
- Devani, Baiju, and Yifan Zhang (2017), “Corporate Bond Markets: Liquidity Determination and Overview,” Investment Industry Regulatory Organization of Canada. 
- Duffie, Darrell, Lasse Heje Pedersen, and Kenneth J. Singleton (2003), “Modeling Sovereign Yield Spreads: A Case Study of Russian Debt,” Journal of Finance, Vol. 58.1, pp. 119-59. 
- Dutta, Prajit K., and Ananth Madhavan, (1997), “Competition and Collusion in Dealer Markets,” Journal of Finance, Vol. 52.1, pp. 245-276. 
- Easley, David, and Maureen O’Hara (1987), “Price, trade size, and information in securities markets,” Journal of Financial Economics, Vol. 19.1, pp. 69-90. 
- Eichengreen, Barry, and Ashok Mody (2004), “Do Collective Action Clauses Raise Borrowing Costs?” The Economic Journal, Vol. 114.495, pp. 247–264. 
- Fang, Chuck, Julian Schumacher, and Christoph Trebesch, Christoph (2021), “Restructuring Sovereign Bonds: Holdouts, Haircuts and the Effectiveness of CACs,” European Central Bank Working Paper Ser. No. 2366. 
- Fang, C., Schumacher, J., and Christian Trebesch (2021), “Restructuring Sovereign Bonds: Holdouts, Haircuts and the Effectiveness of CACs,” IMF Economic Review. 
- Feyen, Erik, Swati Ghosh, Katie Kibuuka, and Subika Farazi (2015), “Global Liquidity and External Bond Issuance in Emerging Markets and Developing Economies,” World Bank Policy Research Working Paper 7363. 
- Fleming, Michael J. (2003), “Measuring Treasury Market Liquidity,” Economic Policy Review, Vol. 9.3, pp. 83-108. 
- Friewald, Nils, Rainer Jankowitsch, and Marti G Subrahmanyam (2012), “Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,” Journal of Financial Economics, Vol. 105.1. pp. 18-36. 
- Gaillard, Norbert (2012), “Fitch, Moody’s, and S&P Sovereign Ratings and EMBI Global Spreads: Lessons from 1993–2007,” in: Gaillard, Norbert (2012), A Century of Sovereign Ratings. New York: Springer, 2012. 
- Garbade, Kenneth, and William Silber (1976), “Price Dispersion in the Government Securities Market,'' Journal of Political Economy, Vol. 84, pp. 721-40. 
- Garbade, Kenneth D. D., and William L. L. Silber (1979), “Structural Organization of Secondary Markets: Clearing Frequency, Dealer Activity and Liquidity Risk,” The Journal of Finance, Vol. 34.3, pp. 577-593.  
- Glosten, Lawrence R, and Paul R Milgrom (1985), “Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders,” Journal of Financial Economics, Vol. 14.1, pp. 71-100. 
- Grosse Steffen, Christoph, Sebastian Grund, and Julia Schumacher (2019), "Collective Action Clauses in the Euro Area: A Law and Economic Analysis of the First Five Years." Capital Markets Law Journal Vol. 14.2, pp. 134-54.  
- Heck, Stephanie, Denis Margaritis, and Aline Muller (2015), “Liquidity Patterns in the U.S. Corporate Bond Market,” 28th Australasian Finance and Banking Conference. 
- Hilscher, Jens, and Yves Nosbusch (2010), “Determinants of Sovereign Risk: Macroeconomic Fundamentals and the Pricing of Sovereign Debt,” Review of Finance, Vol. 14.2, pp. 235-62. 
- Houweling, Patrick , Albert Mentink, and Ton Vorst (2005) “Comparing possible proxies of corporate bond liquidity,” Journal of Banking & Finance, Vol. 29.6, pp. 1331-1358. 
- Hoyos, Ricardo, Yang Liu, Hui Miao, and Christian Saborowski (2020), “Who Drains Bond Market Liquidity in an Emerging Market,” IMF Working Paper WP/20/141. 
- Hund, John, and David A. Lesmond (2008), “Liquidity and Credit Risk in Emerging Debt Markets,” Tulane University. 
- Kalimipalli, Madhu, and Arthur Warga (2001), “Bid/Ask Spread and Volatility in the Corporate Bond Market,” The Journal of Fixed Income, March, pp. 31-43. 
- Ladyman, James, and Karoline Wiesner (2020), What is a Complex System? Yale University Press, USA. 
- Langedijk, Sven, George Monokroussos, and Evangelia Papanagiotou (2017), “Benchmarking Liquidity Proxies: The Case of EU Sovereign Bonds,” European Commission - Joint Research Centre, Ispra, Italy, Vol. 56, pp. 321–329. 
- Lesmond, Da (2005) “Liquidity of Emerging Markets,” Journal of Financial Economics, Vol. 77.2, pp. 411-52. 
- Liu, Qingfu, Renhai Hua, and Yunbi An (2016), “Determinants and Information Content of Intraday Bid-Ask Spreads: Evidence from Chinese Commodity Futures Markets,” Pacific-Basin Finance Journal, Vol. 38, pp. 135–148. 
- Lo, Andrew w., Harry Mamaysky, and Jiang Wang (2004), “Asset Prices and Trading Volume under Fixed Transactions Costs,” Journal of Political Economy, Vol. 112.5, pp. 1054-090. 
- Moutot, Philippe, Imma Curato, and Rafaela Guberovic (2017), “Freeze and Bid-Ask Spread in the Sovereign Bond Market,” SSRN Electronic Journal. 
- O'Hara, Maureen (2003), “Presidential Address: Liquidity and Price Discovery,” The Journal of Finance, Vol. 58.4, pp. vi+1335-1354. 
- Picarelli, Mattia Osvaldo, Aitor Erce, and Xu Jiang (2018), “The Benefits of Reducing Holdout Risk: Evidence from the Euro CAC Experiment, 2013–2018,” Capital Markets Law Journal, Vol. 14.2, pp. 155-77. 
- Ratha, Dilip, Supriyo De, and Sergio Kurlat (2016), “Does Governing Law Affect Bond Spreads?” Policy Research Working Paper No. 7863. World Bank, Washington, DC. 
- Richards, Anthony, and Mark Gugiatti. (2003), “Do Collective Action Clauses Influence Bond Yields? New Evidence from Emerging Markets,” International Finance, Vol. 6.3, pp. 415-47.  
- Rösch, Christoph G, and Christoph Kaserer (2013), “Market Liquidity in the Financial Crisis: The Role of Liquidity Commonality and Flight-to-quality,” Journal of Banking & Finance, Vol. 37.7, pp. 2284-302. 
- Schultz, Paul (2001), “Corporate Bond Trading Costs: A Peek Behind the Curtain,” Journal of Finance 56.2, pp. 677-98. 
- Tanner, J. Ernest, and Levis A. Kochin (1971), “The Determinants of the Difference Between Bid and Ask Prices on Government Bonds,” The Journal of Business, Vol. 44. 4, pp. 375–379. 
- Tsatsaronis, K. (1999), “The effect of collective action clauses on sovereign bond yields.” BIS Quarterly Review: International Banking and Financial Market, November, pp. 22-23. 
- van der Wansen, Patrick, Lars Jessen, and Diego Rivetti (2019), “Issuing International Bonds; A Guidance Note,” MTI Discussion Paper NO. 13, World Bank. 
- Vayanos, Dimitri, and Jean‐Luc Vila, (2021), “A Preferred‐Habitat Model of the Term Structure of Interest Rates,” Econometrica, Vol.89 (1) pp. 77-112. 
- Warga, Arthur (1992), “Bond returns, liquidity, and missing data,” Journal of Financial and Quantitative Analysis, Vol. 27, pp. 605-617. 
- World Bank (2021), International Debt Statistics 2021, Washington D.C. 
- Zhang, Michael Yuanjie, Jeffrey R. Russell, and Ruey S. Tsay (2008), “Determinants of Bid and Ask Quotes and Implications for the Cost of Trading,” Journal of Empirical Finance, Vol. 15.4, pp. 656–678. 

*Sovereign Eurobond Liquidity and Yields Working Paper No. WP/2022/098*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022098-print-pdf.pdf_